Reply information recommendation method and apparatus

1. A reply information recommendation method, comprising: acquiring information to be replied to and received by a user and pre-reply information that is input by the user and corresponding to the information to be replied to; performing segmentation processing on the information to be replied to, to obtain a segmentation processing result; learning a stored text interaction history set of the user to obtain a reply model, wherein the text interaction history set comprises at least one piece of historical information to be replied to that is historically received by the user and reply information corresponding to the at least one piece of historical information to be replied to, and the reply model comprises at least one set of a correspondence between historical information to be replied to and reply information; obtaining candidate reply information with reference to the segmentation processing result of the information to be replied to and the reply model; determining a set of recommended reply information with reference to the candidate reply information and the pre-reply information, wherein the set of recommended reply information comprises multiple expressions having different expression manners at different tone intensities, wherein the set of recommended reply information is determined using semantic matching and fuzzy string matching; and presenting the set of recommended reply information to the user for selection of at least one of the multiple expressions, wherein the set of recommended reply information is presented in an arrangement based on the different expression manners and a ranking of the recommended reply information.

2. The method according to claim 1 , wherein performing segmentation processing on the information to be replied to, to obtain a segmentation processing result comprises: performing, according to a level of a character, a word, a phrase, or a sentence, segmentation on the information to be replied to, to obtain the segmentation processing result.

3. The method according to claim 1 , wherein determining a set of recommended reply information with reference to the candidate reply information and the pre-reply information comprises: generating, according to the candidate reply information, a multi-grammar information set for the information to be replied to, wherein the multi-grammar information set comprises at least one piece of candidate reply information corresponding to the information to be replied to, and a priority of each piece of candidate reply information in the at least one piece of candidate reply information in being used as recommended reply information; combining the pre-reply information and the multi-grammar information set to generate at least one piece of first recommended reply information; performing semantic matching between the pre-reply information and the candidate reply information to generate at least one piece of second recommended reply information; performing fuzzy string matching between the pre-reply information and the candidate reply information to generate at least one piece of third recommended reply information; calculating a conditional probability model according to the multi-grammar information set, wherein the conditional probability model comprises the at least one piece of candidate reply information and use frequency of each piece of candidate reply information in the at least one piece of candidate reply information; and performing, according to the conditional probability model, comprehensive ranking for the first recommended reply information, the second recommended reply information, and the third recommended reply information, to obtain the set of recommended reply information.

4. The method according to claim 3 , wherein calculating a conditional probability model according to the multi-grammar information set, wherein the conditional probability model comprises the at least one piece of candidate reply information and use frequency of each piece of candidate reply information in the at least one piece of candidate reply information comprises: calculating the conditional probability model according to the multi-grammar information set and a stored individual language model of the user, wherein the individual language model comprises a statistical result of reply information that is historically sent by the user.

5. The method according to claim 3 , wherein the first recommended reply information comprises at least one of the following: a recommended character and a recommended word.

6. The method according to claim 3 , wherein the second recommended reply information comprises at least one of the following: a recommended phrase and a recommended sentence.

7. The method according to claim 3 , wherein the third recommended reply information comprises at least one of the following: a recommended phrase and a recommended sentence.

8. The method according to claim 1 , wherein the candidate reply information comprises at least one of the following: a candidate character, a candidate word, a candidate phrase, and a candidate sentence.

9. The method according to claim 1 , wherein the set of recommended reply information comprises at least one of the following: a recommended character, a recommended word, a recommended phrase, and a recommended sentence.

10. A reply information recommendation apparatus, comprising: a memory; and at least one processor configured to: acquire information to be replied to and received by a user and pre-reply information that is input by the user and corresponding to the information to be replied to; perform segmentation processing on the information to be replied to, to obtain a segmentation processing result; learn a stored text interaction history set of the user to obtain a reply model, wherein the text interaction history set comprises at least one piece of historical information to be replied to that is historically received by the user and reply information corresponding to the at least one piece of historical information to be replied to, and the reply model comprises at least one set of a correspondence between historical information to be replied to and reply information; obtain candidate reply information with reference to the segmentation processing result that is of the information to be replied to and the reply model; determine a set of recommended reply information with reference to the candidate reply information and the pre-reply information, wherein the set of recommended reply information comprises multiple expressions having different expression manners at different tone intensities, wherein the set of recommended reply information is determined using semantic matching and fuzzy string matching; and control the apparatus to present the set of recommended reply information to the user for selection of at least one of the multiple expressions, wherein the set of recommended reply information is presented in an arrangement based on the different expression manners and a ranking of the recommended reply information.

11. The apparatus according to claim 10 , wherein the at least one processor is further configured to: perform, according to a level of a character, a word, a phrase, or a sentence, segmentation on the information to be replied to, to obtain the segmentation processing result.

12. The apparatus according to claim 10 , wherein the at least one processor is further configured to: generate, according to the candidate reply information, a multi-grammar information set for the information to be replied to, wherein the multi-grammar information set comprises at least one piece of candidate reply information corresponding to the information to be replied to, and a priority of each piece of candidate reply information in the at least one piece of candidate reply information in being used as recommended reply information; generate at least one piece of first recommended reply information with reference to the pre-reply information and the multi-grammar information set; perform semantic similarity matching between the pre-reply information and the candidate reply information, to generate at least one piece of second recommended reply information; perform fuzzy string matching between the pre-reply information and the candidate reply information, to generate at least one piece of third recommended reply information; calculate a conditional probability model according to the multi-grammar information set, wherein the conditional probability model comprises the at least one piece of candidate reply information and use frequency of each piece of candidate reply information in the at least one piece of candidate reply information; and perform, according to the conditional probability model, comprehensive ranking for the first recommended reply information, the second recommended reply information, and the third recommended reply information, to obtain the set of recommended reply information.

13. The apparatus according to claim 12 , wherein the at least one processor is further configured to: calculate the conditional probability model according to the multi-grammar information set and a stored individual language model, wherein the individual language model comprises a statistical result of reply information that is historically sent by the user.

14. The apparatus according to claim 12 , wherein the first recommended reply information comprises at least one of the following: a recommended character and a recommended word.

15. The apparatus according to claim 12 , wherein the second recommended reply information comprises at least one of the following: a recommended phrase and a recommended sentence.

16. The apparatus according to claim 12 , wherein the third recommended reply information comprises at least one of the following: a recommended phrase and a recommended sentence.

17. The apparatus according to claim 10 , wherein the candidate reply information comprises at least one of the following: a candidate character, a candidate word, a candidate phrase, and a candidate sentence.

18. The apparatus according to claim 10 , wherein the set of recommended reply information comprises at least one of the following: a recommended character, a recommended word, a recommended phrase, and a recommended sentence.